File format & naming

Supported formats

Conflixis will pick up and ingest files with any of these extensions from the configured folder:

| Extension | Notes | | --------- | --------------------------------------------------------------------- | | .csv | Primary format for people and supplier feeds. UTF-8 with BOM is fine. | | .xlsx | Single-sheet Excel workbook; first sheet is used. | | .xls | Legacy Excel; use .xlsx where possible. | | .json | Structured data. Coordinate the schema with your Conflixis contact. | | .txt | Treated as delimited text; coordinate delimiter with your contact. | | .zip | A single-file archive; the inner file must match one of the above. |

CSV is the recommended format for new integrations. It's the most widely tested and the easiest to diff when something looks off.

File naming

• Files are tracked by filename. A filename is only ever imported once. Re-uploading a file with the same name is a no-op.
• To re-send corrected data, change the filename. A date stamp works well — for example, people-2026-04-23.csv.
• Your filePattern glob (for example, *.csv) determines which files Conflixis picks up. Files that don't match are ignored.

Atomic uploads

To prevent Conflixis from reading a file while you're still writing it, upload atomically:

1. Upload to a temporary filename (for example, people-2026-04-23.csv.partial).
2. Rename to the final filename only once the upload completes.

Rename is atomic on most SFTP servers, so this guarantees we only ever see a complete file.

Tip

If your upload tool doesn't support atomic rename, upload into a staging directory that's outside your basePath, then move the file into basePath when complete.

People file (CSV)

For the full list of supported columns, required fields, validation rules, and a worked example, see the dedicated people file schema reference.

Supplier file (CSV)

The supplier feed follows the same pattern. Work with your Conflixis contact to lock down the exact schema; fields typically include supplier name, tax ID, parent/subsidiary relationships, and internal supplier IDs.

CSV rows must not be ragged

Every row in a CSV must have the same number of columns as the header. A file where rows have differing column counts is called ragged, and Conflixis will reject it rather than guess how the fields line up.

• Each data row must contain exactly as many delimited fields as the header row — no more, no fewer.
• A field that has no value must still be present as an empty field (for example, Jane,,jane@example.com), not omitted.
• If a field's value legitimately contains the delimiter, a quote, or a line break, wrap the whole field in double quotes and escape embedded quotes by doubling them ("Acme, Inc.", "She said ""hi"""). This keeps the field count stable across every row.
• Don't pad rows with extra trailing delimiters, and don't add commentary, totals, or blank separator lines between data rows.

Caution

Trailing empty columns are the most common cause of raggedness — a tool that exports a "notes" column only when it's populated will silently drop the field on other rows. Export every column on every row, even when empty.

Encoding and line endings

• UTF-8 is required. UTF-8 with or without a BOM is fine.
• Both LF and CRLF line endings are accepted.
• Avoid Windows-1252 / Latin-1 encoded files — non-ASCII characters will be corrupted.

What to do about PII

The people feed typically contains personally identifiable information (names, work emails, sometimes manager relationships). Treat the files the same way you'd treat any export from your HRIS:

• Use the dedicated conflixis user and a directory that isn't browsable by other users on the SFTP server.
• Rotate credentials (keys or passwords) on your normal cadence.
• If your organization requires encryption-at-rest on the SFTP server, verify that your server provides it.

Conflixis encrypts the file in transit (SFTP), at rest in our object store, and in our application database.